The neuronal SNARE complex is at the heart of neuronal exocytosis. It is an extended coiled coil consisting of four alpha-helical SNARE-motifs, provided by syntaxin-1, synaptobrevin-2 and SNAP-25. The formation energy of the complex is assumed to overcome the energy barrier for fusion, but it has been unclear during which phases leading up to neuronal exocytosis the SNARE-complex forms. Here, I present recent data on the role of the SNARE-complex in vesicle docking, spontaneous release and fusion pore formation. Experiments in chromaffin cells show that two SNARE-complex members, syntaxin-1 and SNAP-25, but not synaptobrevin-2, are involved in docking vesicles to the plasma membrane. The vesicular counterpart is synaptotagmin-1, better known as the calcium censor for exocytosis, which binds to an 1:1 SNAP25:syntaxin acceptor complex on the plasma membrane. This docking reaction does not require Munc18-1, but Munc18-1 plays an indirect role, by stabilizing acceptor complexes and stimulate SNARE-complex assembly. Next, the N-terminal end of the synaptobrevin-2 SNARE-motif joins the complex, driving vesicle priming, the reaction that makes the vesicle releasable. Fusion triggering is executed by calcium binding to synaptotagmin, which causes the C-terminal end of synaptobrevin-2 to 'zipper up' to the remaining complex. This part of the reaction leads to rapid fusion, and to fusion pore formation. Thus, we find that C- but not N-terminal mutations in synaptobrevin-1 change fusion pore duration. The last piece of evidence concerns the origin of spontaneous release events in neurons, which have been attributed to another fusion machinery, or to separate vesicle pools. Recent data obtained in autaptic neurons indicate that spontaneous release events can be inhibited or stimulated by C- and N-terminal mutations in the SNARE-complex, respectively. This indicates that spontaneous release events are driven by the same fundamental machinery as evoked release. The SNARE-complex helps shape the energy landscape for fusion so as to favour evoked release. In sum, our data allow the synthesis of a model for vesicle fusion from the initial docking of vesicles to the plasma membrane until the formation of a fusion pore.